Intellectual Property India Publishes Patent Application for Deep Learning-Based Detection Of Chest Abnormalities From Medical Imaging

MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202611057062 A) filed by Divyanshu Pathak; Ananya Srivastava; Aneesh Kumar Shahi; Atul Kumar Kaushik; Shivangi Tyagi; Neha Bhatia; Manvi Khatri; Umang Kant; Anjali Chauhan; Ashima Arya; and Mayank Tyagi on May 05, 2026, for Deep Learning-Based Detection Of Chest Abnormalities From Medical Imaging.

Inventors include Divyanshu Pathak; Ananya Srivastava; Aneesh Kumar Shahi; Atul Kumar Kaushik; Shivangi Tyagi; Neha Bhatia; Manvi Khatri; Umang Kant; Anjali Chauhan; Ashima Arya; and Mayank Tyagi.

The application for the patent was published on July 17, 2026, under issue no. 29/2026.

Abstract: The present invention discloses an automated multi-label chest abnormality detection system (100) that integrates a plurality of transfer-learning-based deep convolutional neural network models with a Multi-Criteria Decision-Making (MCDM) evaluation framework for objective model selection. The system employs pre-processing including multi-hot label encoding for fourteen thoracic disease categories, class imbalance resampling, and data augmentation applied to the NIH ChestX-ray14 dataset. Six deep learning architectures including MobileNet (110), ResNet50 (120), VGG16 (130), DenseNet201 (140), Xception (150), and a Hyperparameter- Tuned Xception model (160) are trained with sigmoid-activated multi-label output layers. The MCDM evaluation module (170) employs Weighted Sum Method, TOPSIS, VIKOR, and Rank Position Method to rank models across weighted performance metrics, consistently identifying the Hyperparameter-Tuned Xception model (160) as the optimal clinical inference model. The system provides a clinical decision support interface (180) for presentation of per-class disease probabilities to support radiologists.

Disclaimer: Curated by HT Syndication.